Artwork

Kandungan disediakan oleh TWIML and Sam Charrington. Semua kandungan podcast termasuk episod, grafik dan perihalan podcast dimuat naik dan disediakan terus oleh TWIML and Sam Charrington atau rakan kongsi platform podcast mereka. Jika anda percaya seseorang menggunakan karya berhak cipta anda tanpa kebenaran anda, anda boleh mengikuti proses yang digariskan di sini https://ms.player.fm/legal.
Player FM - Aplikasi Podcast
Pergi ke luar talian dengan aplikasi Player FM !

Localizing and Editing Knowledge in LLMs with Peter Hase - #679

49:46
 
Kongsi
 

Manage episode 411383461 series 2355587
Kandungan disediakan oleh TWIML and Sam Charrington. Semua kandungan podcast termasuk episod, grafik dan perihalan podcast dimuat naik dan disediakan terus oleh TWIML and Sam Charrington atau rakan kongsi platform podcast mereka. Jika anda percaya seseorang menggunakan karya berhak cipta anda tanpa kebenaran anda, anda boleh mengikuti proses yang digariskan di sini https://ms.player.fm/legal.

Today we're joined by Peter Hase, a fifth-year PhD student at the University of North Carolina NLP lab. We discuss "scalable oversight", and the importance of developing a deeper understanding of how large neural networks make decisions. We learn how matrices are probed by interpretability researchers, and explore the two schools of thought regarding how LLMs store knowledge. Finally, we discuss the importance of deleting sensitive information from model weights, and how "easy-to-hard generalization" could increase the risk of releasing open-source foundation models.

The complete show notes for this episode can be found at twimlai.com/go/679.

  continue reading

701 episod

Artwork
iconKongsi
 
Manage episode 411383461 series 2355587
Kandungan disediakan oleh TWIML and Sam Charrington. Semua kandungan podcast termasuk episod, grafik dan perihalan podcast dimuat naik dan disediakan terus oleh TWIML and Sam Charrington atau rakan kongsi platform podcast mereka. Jika anda percaya seseorang menggunakan karya berhak cipta anda tanpa kebenaran anda, anda boleh mengikuti proses yang digariskan di sini https://ms.player.fm/legal.

Today we're joined by Peter Hase, a fifth-year PhD student at the University of North Carolina NLP lab. We discuss "scalable oversight", and the importance of developing a deeper understanding of how large neural networks make decisions. We learn how matrices are probed by interpretability researchers, and explore the two schools of thought regarding how LLMs store knowledge. Finally, we discuss the importance of deleting sensitive information from model weights, and how "easy-to-hard generalization" could increase the risk of releasing open-source foundation models.

The complete show notes for this episode can be found at twimlai.com/go/679.

  continue reading

701 episod

Semua episod

×
 
Loading …

Selamat datang ke Player FM

Player FM mengimbas laman-laman web bagi podcast berkualiti tinggi untuk anda nikmati sekarang. Ia merupakan aplikasi podcast terbaik dan berfungsi untuk Android, iPhone, dan web. Daftar untuk melaraskan langganan merentasi peranti.

 

Panduan Rujukan Pantas

Podcast Teratas